Strategies

Lynx Constellation Program

The Lynx Constellation Program's objective is to generate attractive risk adjusted returns which are differentiated from other traditional and alternative investments by employing machine learning models to identify both linear and non-linear relationships across a broadly diversified portfolio of markets.

We won an award!

We are proud to announce that the Lynx Constellation Program won the Best CTA award for 2023 at the Hedgeweek European Awards!

For full details on nominations and selection criteria, please click here.

Lynx was early in applying machine learning techniques in finance. We hired our first machine learning expert in 2009 and today these advanced techniques have grown to be an important part of our model portfolio. In the video below you can find out more about what machine learning means to us and how we apply it in our investment approach.

Experienced – Machine learning models have been employed in the Lynx Program since 2011 and have historically generated solidly positive results

Diversified – Trades approximately 100 markets using signals generated by a diverse group of models spanning multiple timeframes and machine learning concepts

Adaptable – Models employ machine learning techniques designed to learn from – and adapt to – different market environments

Historical performance

Lynx Constellation
Choose index

Investing in funds is associated with risk. Past performance is no guarantee of future return. The numbers shown above should be read together with this clarifying note.

Lynx Constellation

Total return

Average annual return

Sharpe ratio

Maximum drawdown

Standard deviation

Lynx Constellation

Total return

Average annual return

Sharpe ratio

Maximum drawdown

Standard deviation

Investment strategy

Lynx Constellation employs systematic models utilizing a range of machine learning techniques to forecast market prices. These forecasts are assigned dynamic weights based on their expected prediction accuracy and marginal contribution to the portfolio. An optimal portfolio is then constructed attempting to maximize risk-adjusted return while minimizing trading costs. The forecasting models are equipped to identify and exploit imbalances in the most liquid futures markets globally. Some of these imbalances are based on investor tendencies and behavioral biases such as herding, while others are based on repeating patterns of price action influenced by other markets and/or factors such as seasonality. As the models constantly adapt to – and learn from – new information, the market phenomenon exploited at any given time will change with the environment.

Risk management and risk considerations

Risk management in Lynx Constellation is based on three pillars: model-driven risk control, robust portfolio construction and a top-down risk limit framework. The models incorporate risk management elements into the signal generation process, increasing or decreasing risk based on size constraints and volatility. During the portfolio construction process, the optimizer considers volatility and correlation between markets when determining position sizes and total portfolio risk. Finally, we utilize a risk limit framework on a market and portfolio level, employing three separate models in parallel based on Value at Risk to measure market risk over different time frames; the model stating the highest measure at any given time is used to limit risk. Investing in funds is associated with risk. Past performance is no guarantee of future return. The value of the capital invested in the fund may increase or decrease and investors cannot be certain of recovering all of their invested capital.

How to invest

The Lynx Constellation managed accounts or offshore vehicles are intended for institutional investors. If you want more information or want to to discuss a solution that may suit your needs please contact our investor relations team.

Approach

Rigor and innovation are key to developing our systematic models that are tasked with analyzing and acting on vast amounts of information.